PROCESS MONITORING AND DIAGNOSIS BY MULTIBLOCK PLS METHODS

Schemes for monitoring the operating performance of large continuous p rocesses using multivariate statistical projection methods such as pri ncipal component analysis (PCA) and projection to latent structures (P LS) are extended to situations where the processes can be naturally bl ocked in to subsections. The multiblock projection methods allow one t o establish monitoring charts for the individual process subsections a s well as for the en tire process. When a special event or fault occur s in a subsection of the process, these multiblock methods can general ly detect the event earlier and reveal the subsection within which the event has occurred. More detailed diagnostic methods based on interro gating the underlying PCA/PLS models are also developed. These methods show those process variables which are the main contributors to any d eviations that have occurred, thereby allowing one to diagnose the cau se of the event more easily. These ideas are demonstrated using detail ed simulation studies on a multisection tubular reactor for the produc tion of low-density polyethylene.